Incident response is a skill that cannot be learned from documentation, and almost all incident response training is documentation. This is the central problem with the category and it is worth being precise about why. The knowledge component of incident response is not large. Declare early.
Establish a single incident commander. Separate the person fixing from the person communicating. Keep a timeline. Mitigate before diagnosing.
Do not let a side investigation capture the main channel. You can read that list in four minutes and most on-call engineers have. The difficulty is not knowing it, it is executing it at three in the morning when you have been woken from sleep, the dashboard is showing four things wrong at once, a senior person has joined the channel and is asking for an ETA, and your working memory is substantially degraded by adrenaline and interrupted sleep. Under those conditions people do not follow the process they read about.
They revert to whatever they have practised. Rootly Academy is built on that observation, and it is the right observation. The format is drills: short scenarios, ten to thirty minutes, in which a realistic incident unfolds against real tools, with a fictive team producing the messages and requests that a real team would produce. You are not watching someone describe an incident.
You are in one, with the clock running, having to decide what to look at first and what to tell people. The partnership with Uptime Labs matters here, because Uptime Labs has been building this kind of simulation longer than Rootly has been building incident tooling, and the difference between a well-designed drill and a dressed-up quiz is entirely in the design. The drill length is a genuinely good decision that sounds trivial. Training that requires a half-day block does not get done, because the half day never arrives.
Twenty minutes fits in the gap after standup, which is the difference between a programme people complete and a programme people intend to complete. The scenario coverage extending beyond technical outages into security incidents and communication failures is also correct, because the incidents that damage organisations most are rarely the ones where the technical diagnosis was hard. They are the ones where the diagnosis was fine and the coordination was a disaster: three people investigating the same hypothesis, nobody talking to customers for ninety minutes, a status page updated after the journalists called. Practising the coordination is practising the thing that actually fails.
My criticisms are mostly about commercial opacity and newness. Pricing is not published anywhere and the site routes you to booking a demo, which is the enterprise sales pattern and is a genuine barrier for exactly the audience that would benefit most: the individual engineer who wants to get better at this, or the team lead with a small training budget and no appetite for a procurement conversation. Compare this with PagerDuty, which open sourced its internal incident response documentation at response.pagerduty.com with no login, no gate and no sales call, and which I have described elsewhere in this catalogue as the best free incident response material available. Rootly Academy is a better format than PagerDuty's documentation.
PagerDuty's is infinitely more accessible. Use both: read the doctrine free, then practise it if you can get the budget. The programme launched at SREcon North America in March 2026, which makes it new. New is not a criticism in itself, but a drill library's value is proportional to its breadth, and a catalogue that is still building will have gaps.
Ask what exists before you commit a team to it. The tool-agnosticism question is the one I would want answered most clearly. The drills use real tools including Grafana, Slack and Rootly, and the honest question is how much of what you practise is portable incident response skill and how much is familiarity with Rootly's interface. I suspect the answer is mostly the former, because the coordination and decision-making content is genuinely product-independent, but a buyer should ask rather than assume.
The structural point I make about every vendor academy applies here too, though more gently than usual. Rootly sells incident management software, and a company that has fewer incidents needs less incident management software. The material is therefore never going to push hard on the upstream question, which is what your change management, testing discipline, deployment strategy and architectural choices are doing to your incident rate. Responding well to a hundred incidents a year is a worse outcome than responding adequately to twenty, and no incident response vendor will structure a curriculum around that.
To be fair to Rootly, this is a milder conflict than the one I have noted for observability vendors billing by data volume, and the response skills being taught are genuinely needed regardless. One honest caveat about simulation itself, which the format's advocates tend to underplay. A drill you have opted into during working hours, knowing it is a drill, does not reproduce the physiological state of a real page at four in the morning. It builds the procedural memory, which is the majority of the benefit, but the claim that it prepares you for the real thing should be held a little loosely.
The nearest thing to real practice remains a well-run game day against production-like systems with real stakes, and this complements that rather than replacing it. Four point three. The right format, well constructed, with a partner who knows what they are doing, held back by opaque pricing and a young content library rather than by anything wrong with the approach.